Exploring the Potential of Generative Adversarial Networks in Enhancing Urban Renewal Efficiency

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The authors explored how the established GAN framework can be applied into generating architectural forms and structures via image generation techniques, which is an important contribution towards sustainability development and urban analytics. However, there are some potential room for improvement based on the existing manuscript:

Major Points to be addressed:

(1) Introduction: The authors have highlighted the use of GANs and an extended GAN-based algorithm named pix2pix for image processing and generation. Please try to take more references of recent usage of GAN-based models for data clustering and image generation. Some references are as follows:

(a) https://www.mdpi.com/1424-8220/23/7/3457 (in generating individual game levels and combine the game levels via image processing techniques)

(b) https://link.springer.com/article/10.1007/s00500-022-07026-7 (in classification of clustered health related datasets)

(c) https://www.sciencedirect.com/science/article/pii/S1077314222001497 (in understanding the structure and features of images and for computer vision)

(2) For the pix2pix algorithm, is there any data preprocessing stage? If so, please describe it. If not, please explain briefly why it is not needed in this particular scenario.

(3) Section 2.1: For the pix2pix algorithm, why does it considering taking L1 norm rather than other mathematical norms?

(4) Section 3.1 (Lines 209-214): The idea proposed and conducted here is good from machine learning perspectives. But are these geometric / physical transformations realistic? Any proper reference for making such changes?

(5) Section 3.2.2: The iteration counts are now taken as 100, 200, 300 and 400. Is there any stopping criterion? What if in practice, the convergence is not achieved after 400 iterations are conducted?

(6) Section 3.3: In Lines 268-269, the authors mentioned that "pix2pix demonstrates the capacity to generate reasonably plausible alternations" - is there any statistical metric to support the claim? Please provide some other quantitative results apart from the ones shown in Table 2.

(7) In addition to point (6) above, in Section 3.3.2, the authors should provide the correlation and percentage of matching of image generated by pix2pix algorithm, so that the manuscript will look more scientific and persuasive.

(8) Section 3.3.3: The use of PSNR and SSIM are great, but may not be sufficient. Please also consider the use of time complexity and average loss during pix2pix training, and include some of these numerical descriptions in your manuscript.

(9) In Conclusion, the potential extension of this study is missing, it's too short for now. Please enrich its content accordingly.

Minor Points to be addressed:

(1) The comparison was conducted against a test set. Is the test set reliable, how is the reliability ensured? Any official source of the images used for intercomparison or validation?

(2) In Introduction, please define sustainability and urbanization first. In particular, please explain how the architectural design features are related to urban sustainability. Please provide some literature showing the quantitative relationships.

(3) Line 191: What parameters of the discriminator network are updated?

(4) Line 220: "lead to a certain discount" - please quantify such discount, and provide some explanations behind the claim?

(5) Line 291: For "pixel distributions within images", do you refer to intensity or darkness or types?

(6) Figure 6: Please provide the numerical and qualitative differences when 100 and 200 iterations are used for generating images.

(7) Table 2: For PSNR (average), what is said to be a reasonably good numerical figure?

(8) Line 383: How do you define "solution efficiency"? In terms of time complexity of the algorithm or?

(9) Lines 389-390: How do you conduct further optimization from architectural discipline? Please suggest some precise ways.

Overall speaking, the manuscript is creative and is practical, but it should be written in a more scientific manner, and I think that the manuscript is a good one after the authors addressed the aforementioned comments.

Comments on the Quality of English Language

The English writing and grammatical patterns of sentences are mostly fine, except with minor error. It looks easy to follow.

Line 142: and possess

Line 147: we summarize the findings of this study in Section 5.

Line 152: attempts

Line 217: aims at enabling

Line 272: match with the target image

Line 381: images when compared to the original ones

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

The reviewed article addresses the interesting issue of attempting to apply GAN networks to the renovation of urban factories using the example of a dilapidated industrial park in Changsha, China. The presented research is part of the worldwide trend of increasing use of machine learning and artificial intelligence in labor- and time-intensive processes and tasks.

The very idea/idea of using advanced technologies and digital programs to perform tasks previously performed only by humans should be considered a strength of the article. As the study showed, the proposed methodology of proceeding, allowed to achieve the intended results with a very high degree of accuracy and precision in a much shorter time and with fewer resources.

However, it should be noted that in the article there is no clear answer as to whether the task performed by the computer contains the same aesthetic and artistic qualities as before, when they were performed by architects. It would be worthwhile to add such a narrative thread in the presented research (just because a computer is able to do something does not mean that it will also be of good quality).

In summary, the article presented for review is part of a worldwide research trend describing the use of increasingly complex digital tools and programs in human activities. The article is written correctly and presents a clear research methodology and results. With a little proofreading, it is suitable for publication.

Author Response

Please see the attachment

Author Response File: Author Response.docx

Reviewer 3 Report

Comments and Suggestions for Authors

In the manuscript titled Exploring the Potential of Generative Adversarial Networks in Enhancing Urban Renewal Efficiency, Yunfei Lin and Mingxing Song evaluated the role of pertinent technologies in enhancing the effectiveness of factory refurbishment in the framework of the revitalization of historic factory buildings in industrial regions, hence advancing the development of urban regeneration in the future. With the models generated by this study, factory profiles based on design features will have more options for future designers to choose from when addressing issues.

 

On the subject, the article offers fresh facts. Although the writers have studied the literature pertinent to the topic under discussion. I believe their thoughts have not been sufficiently explained. Although the article's structure is stable, condensing the language might improve it. The research questions' answers were poorly articulated throughout the debate, and while the discussion of the pertinent subjects was summarized, no summative review took place. To improve the experiment's usefulness in the future, the authors could offer more detailed instructions for using the technique.

 

Authors need to make changes based on suggestions.

Author Response

Please see the attachment

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

The authors have addressed all my aforementioned comments. In terms of science and logic of the manuscript, it is well done and completed now. There is just a minor comment to go before publication:

For Introduction, the authors mentioned sustainable resource utilization, revitalization, enhancing urban renewal efficiency and urban development etc. Perhaps it would be much better if they could also include liveability conditions, effective land utilization for city development, urban architectural design and smart urban governance in their manuscript (can be incorporated into various parts). Some references are as follows:

https://www.mdpi.com/2071-1050/13/16/8781

https://www.mdpi.com/2073-445X/12/4/894

https://www.sciencedirect.com/science/article/abs/pii/S0264275120313524

https://www.epfl.ch/labs/lasur/wp-content/uploads/2018/05/SAUER.pdf

After the authors incorporate these ideas into the revised manuscript, I think it's good to be published.

 

Comments on the Quality of English Language

It's good. Minor grammatical checking is encouraged.

Author Response

Please see the attachment

Author Response File: Author Response.pdf